本文以美國、德國、日本等國之股價指數與指數期貨為主要研究對象,美國史坦普500股價指數、德國法蘭克福指數、日本日經225指數研究期間取自1991年1月1日至2009年12月31日止,美國道瓊工業指數研究期間取自1998年1月1日至2009年12月31日止。運用不同避險績效的衡量方法,包括變異數(Variance) 、效用函數(Utility function)、半變異數(semi-variance)、低度動差(LPM)、條件風險值(CVaR)等來估計OLS、CCC-GARCH、DCC-GARCH、DCC-CARR等避險模型之樣本外避險績效。實證結果發現:1.觀察樣本外期間的避險績效,以OLS避險模型效果最佳。2.若僅比較DCC-GARCH及DCC-CARR模型,以DCC-CARR 模型做為波動性預測指標的動態模型,比DCC-GARCH 模型做為波動性預測指標的動態模型估計更準確。3.本文同時考慮避險交易成本之因素,運用效用函數來評估四大指數避險績效,以OLS模型及DCC-CARR模型避險績效為佳,而考慮交易成本後之DCC-CARR模型避險績效同樣亦較DCC-GARCH模型的避險效果為佳。
This article takes stock index and index future in United States, Germany and Japan as the research object . The sample period of S&P 500、DAX and Nikkei 225 index covers from 1/1/1991 to 31/12/2009, and the sample period of Dow Jones index covers from 1/1/1998 to 12/31/2009. The purpose of this study is to compare the out of sample performances among OLS、CCC-GARCH、DCC-GARCH、DCC-CARR models by using Variance、Utility function、Semi-variance、LPM and CVaR measurements. The empirical result shows: 1. OLS hedging model has the best out of sample performance. 2. In the dynamic model, if it only compares DCC-GARCH and DCC-CARR as the volatility forecasting estimator, DCC-CARR model has more accurate result. 3. If it takes transaction costs into the consideration, Utility function shows OLS and DCC-CARR models both have better hedging performances. Consideration the transaction cost, DCC-CARR model is also better than DCC-GARCH model.